Safebook CEO Says AI Agents Solve CFO’s Data Plumbing Problem

Corporate finance has a paradox problem. Businesses move faster than ever, yet the teams responsible for tracking, validating and reporting the money still operate as if speed were optional. While sales, marketing and customer support have spent years layering automation and AI onto their workflows, finance remains anchored to spreadsheets, manual reconciliation and systems that were never designed to work together.

Even as 2026 approaches, the office of the CFO is still defined by labor-intensive processes and fragmented technology. Unlike more automated enterprise functions, finance teams continue to stitch together data from ERPs, CRMs, billing platforms, contract repositories and banking systems — and often by hand. The result is slow, error-prone workflows that strain teams just as expectations for real-time insight keep rising.

That widening gap between business velocity and financial reality is what led Safebooks to raise a $15 million seed round and emerge from stealth December 9. The company is betting that AI agents can finally solve what Ahikam Kaufman, co-founder and CEO of Safebooks, calls the CFO’s “data plumbing” problem. That will free accountants from verification work and allow them to focus on judgment-based decisions that actually require human expertise.

Kaufman sees finance at an inflection point. During a discussion hosted by PYMNTS CEO Karen Webster, he described what he hears repeatedly from finance leaders: frustration that transformation has passed them by. There is no shortage of software in finance, he argued, if anything, there may be too much. Each system solves a specific problem, but together they create a fragmented data landscape that accountants must constantly reconcile, toggling between systems and spreadsheets.

Modern finance organizations operate across ERPs, CRMs, CPQ tools, billing platforms, banking systems and document repositories. Each, Kaufmann explains, holds a partial version of the truth, often using different naming conventions and update cadences. Accountants are left to connect the dots manually, validating data across systems before they can even begin making accounting decisions.

Safebooks is designed to sit on top of those existing systems, unifying finance data to enable a faster, more accurate Time to Cash. Webster described the platform as a way to automate reconciliation, reduce risk and deliver real-time insight without forcing finance teams to grow headcount. The company’s recent funding round reflects investor confidence that this approach can finally modernize order-to-cash processes at scale.

At the heart of the problem, Kaufman explained, finance teams are forced into two very different kinds of work. One involves validating data integrity across fragmented systems. That’s checking that contracts match CRM records, billing entries and ERP data. The other involves making accounting decisions, such as when to recognize revenue or whether to approve a deal. Only the latter truly benefits from human judgment.

Verifying data, by contrast, is pure plumbing work. It is repetitive, time-consuming and increasingly complex as transaction volumes grow. What was once contained within a single ERP now spans dozens of systems. Over time this fragmentation has compounded, turning routine finance operations into a constant exercise in reconciliation.

Safebooks approaches the problem by reading both structured and unstructured data. Its platform ingests contracts, order forms, CRM records, billing entries and ERP data, then maps everything into a unified financial data graph. Once that unified data set exists, the system can automatically identify anomalies, inconsistencies, reconciliation issues and discrepancies across thousands of transactions.

The impact, Kaufman said, can be dramatic. One customer reduced contract processing time from roughly 22 minutes to just 22 seconds. That kind of acceleration matters when finance teams are dealing with thousands of transactions each month and are expected to deliver real-time answers about what is in the bank, what is in process and what is expected.

Webster pressed Kaufman on one of the most persistent challenges in finance automation: timing mismatches. Some systems update instantly, while others, particularly ERPs, operate in batches. In revenue operations, those delays can create exposure long before an issue is formally booked.

Kaufman argued that this is precisely why Safebooks focuses on order-to-cash. It is the most sensitive area from a compliance standpoint and the most visible to customers. Errors don’t stay hidden; customers can catch them. In enterprise environments, there may be fewer transactions than payments, but far more data points must be checked constantly for compliance, leakage and billing accuracy.

At scale, manual oversight simply breaks down. Kaufman noted that there is no human capability to read through every system for 5,000 revenue transactions and verify multiple data points per transaction. The complexity becomes unmanageable. Even with unlimited staff, manually overseeing tens or hundreds of billions of dollars in monthly payment volume is not realistic.

AI agents change that equation by allowing finance teams to scale output without scaling headcount. Automation runs continuously. It does not get tired, take vacations or batch work into weekly reviews. Instead, AI agents can read contracts, cross-check systems and flag issues as they occur, reducing both operational risk and compliance exposure.

That capability becomes especially important for companies during periods of rapid growth. Kaufman points to tech companies that are growing 25%, 50%, or even 100% year over year, and that quickly discover that manual processes do not scale. The default response has traditionally been to hire more people just to keep up. Safebooks aims to offer an alternative — one where accuracy, speed and control improve even as transaction volumes explode.

Once finance teams experience that shift, Kaufman believes there is no going back.  Exposure to automation changes expectations permanently. He summed it up with a lesson he attributes to Steve Jobs.

The real breakthrough is not replacing humans, but combining human judgment with powerful tools. In finance, that combination may finally allow the office of the CFO to move at the speed modern businesses demand.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.